company

Comparing R, PAL, and APL Capabilities
Because SAP HANA supports multiple execution
engines, users can choose the best predictive
library for the task – and even combine usage of
multiple libraries at once. It is important to realize
that R, PAL, and APL are complementary technologies
to provide a choice in how to plan out projects.
The following sections discuss the decision
criteria for choosing which library or combination
of libraries would be most appropriate for a specific
use case.
Note: SAP recommends deploying APL and PAL
in all SAP HANA platform deployments that
require predictive functionality. Customers who
need additional functionality beyond what PAL
provides can also deploy an R server in a sidecar
configuration.
ALGORITHMIC VERSUS AUTOMATED
PREDICTIVE: R AND PAL VERSUS APL
The choice between algorithmic (R and PAL) and
automated (APL) predictive capabilities largely is
dependent on the target users and their needs.
The APL provides flexibility to automate the predictive
analytics workflow without users needing
knowledge of how to build complex data models
from scratch. PAL or R typically requires a user to
create procedures manually for each stage of the
predictive modeling workflow.
Data scientists naturally tend to prefer algorithmic
techniques that offer a high degree of control
and precision in the modeling process. This flexibility
comes at a cost: both R and PAL require
users to be properly trained in data science techniques,
as they must understand what each algorithm
does, how it works, and how to interpret
the results. Even seasoned data scientists working
on less-sophisticated analyses need to invest
time to go through the full predictive analytics
workflow on each problem when using algorithmic
techniques.
Automated analytics automates many of the
predictive-modeling steps that a data scientist
typically performs for common workflows like
classification, regression, and association analysis,
saving the user time and effort. The automated
machine learning engine still performs the
full predictive analytics workflow but requires very
little input from the user. The result is a significantly
faster analysis that has fewer configuration
parameters.
In general, both data scientists and business
analysts should start their analysis by using the
automated predictive capabilities of SAP HANA
whenever possible. Automated machine learning
can address a growing number of scenarios and
typically can produce valid results in seconds or
minutes. This enables those who are not data
scientists to answer their own questions and
quickly iterate on the results in a self-service
manner while giving data scientists an automated
way of analyzing many problems quickly.
In some cases where a data scientist may want
to create a more complex model or be in complete
control of each algorithmic parameter, it
is appropriate still to start with APL. That way,
the data scientist can understand the data and
create hypotheses before transitioning to an
algorithmic method such as SAP HANA–native
PAL or offboard R scripts.
© 2016 SAP SE or an SAP affiliate company. All rights reserved.
31 / 33